Transcription
[Music] As you know, psychology is a study of the scientific study of behavior and mental processes. And environmental psychology is no different from other areas of psychology in that it's largely a research-based field. But as I pointed out last time, unlike other areas of psychology, environmental psychologists are more likely to use applied research methods and less likely to do laboratory experiments; they do laboratory experiments sometimes, though. So let's talk a little bit about the main features of laboratory experiments along with an example.
[Music] Now, as you probably remember from a past psychology class, laboratory experiments have three main features: (1) Manipulation of an independent variable; (2) Random assignment to independent variable conditions; and (3) Experimental control. So here's an example. Last time I talked a little bit about some noise studies, and I mentioned that those studies find that predictability of noise and the amount of control you have over the noise affect how stressful it is. That idea comes originally from some early studies in the late 1960s and the early 1970s by Glass and Singer. The independent variables that they manipulated in those studies were whether the noise was unpredictable or predictable, and whether participants thought they had control over the noise—that is, the power to stop it. The noise usually lasted between 20 and 25 minutes, depending on the particular experiment. Now this is a controlled lab setting—because that's key to experiments. And so the white male participants are randomly assigned to the different IV noise conditions, but otherwise they're treated identically, and everything is held constant—that is, it's standardized across the groups—so the only thing different between them is going to be the type of noise that they were exposed to. That way, if the groups differ on the effect variable—what we call the dependent variable—then the researchers would know it was due to the independent variable. That is, they can conclude there's a cause-effect relationship.
Now the dependent variable. Remember, this is the effect variable in your cause-effect hypothesis. Here, DVs included physiological measures of stress, frustration with noise, and performance on both simple and complex tasks. So what they did basically was compare the groups that differed on the independent variables on the dependent variables. So I'm going to simplify here, but generally speaking, they found that unpredictable, uncontrollable noise has the most negative effects; there are bigger effects on the performance of complex tasks; and the psychological effects of noise can linger.
Now laboratory experiments like this are considered to be high in internal validity, and what that means is that the experimenters can confidently conclude from the study results whether or not there's a cause-effect relationship between the independent and dependent variables; so that's a good thing, and this is one reason why experiments are so popular as a research method. So the point of both random assignment and experimental control is to make sure that the independent variable groups are identical except for the independent variable; remember, different groups get different versions or levels of the independent variable. Random assignment, for instance, makes sure that the groups are comprised of the same types of people—that is, that education, experience, etc., are evenly distributed across all IV groups such that they end up being the same as far as the kinds of people in them. Meanwhile, experimental control means they're treated the same, they're tested at the same time of day, use the same experimenter. You even standardize the instructions, and all those things make sure that the only difference between the groups is the independent variable. And the idea is that if the groups are identical except for the independent variable, then if they differ on the dependent variable, you know it was due to the independent variable, and you can conclude there's a cause-effect relationship. And that is what makes experiments high in internal validity.
Unfortunately, this is also what makes lab experiments low in external validity. External validity is the ability to generalize the results to real-world settings and to a diverse array of people. So remember, this is an artificial lab environment where the noise is stimulated, people know they're in an experiment, and they're using white male college students, etc. You know, knowing you're in an experiment often affects your behavior. People tend to exhibit the behaviors that they think are socially desirable. Unlike atoms or rats, people know they're being studied, and that can affect their behavior. Another problem: Sometimes the way things are represented in the experiment are "fake." Concepts, like noise in this case, differ in some important ways from the way they're experienced in real life. So yeah, lab experiments tend to be very high in internal validity but lower in external validity.
[Music] Field experiments are conducted in real-world settings, and in some ways they offer the best of both the internal and external validity worlds. They manipulate independent variables—IVs as we call them for short. They use random assignment to IV conditions, and they control for extraneous factors through the use of control groups. So, for example, when I was a young pup, a young graduate school pup in my early 20s in the early 1980s, I conducted a field experiment. I wanted to test some low-cost persuasive methods for increasing participation in one of the country's first curbside recycling programs. I randomly assigned small groups of non-recycling homes to one of three independent variable groups or a control group. I trained Boy Scouts to knock on the doors, give a short description of the program, and administer the IV treatments. One treatment was a persuasive communication based on laboratory research on persuasion. The homeowner was asked to read the statement on the spot. Another treatment was a card representing a public commitment persuasive technique. The homeowner was asked to sign the card expressing support for the city's recycling program. A third treatment combined the persuasive communication and the public commitment. There was also a no-treatment control group that had no contact with the Scouts. Participation in the curbside recycling program was measured for six weeks afterward by driving around and recording whether homeowners placed their recyclables at the curb on trash collection day. Basically, I found that the treatment groups differed significantly from the no-treatment control condition but not from each other. Approximately 44% of the households in the treatment conditions started recycling in contrast to 11% of the control group.
Now this is a field experiment done in a real-world setting, and it's high in internal validity because I randomly assigned homes to the treatments to ensure that the different IV groups were as similar as possible except for the IV. There was also a no-treatment control group which controlled for what's called a "history threat to validity"—that would be if something else happened during the study period that was responsible for increasing the recycling. The control group controls for that possibility. That 11% increase in that group compared to the 44 percent increase represents other things that increased recycling at that time. The study is relatively high in external validity because it was a field study conducted in the real world, in a real place using real "indigenous personnel" (Scouts), so the techniques could be easily used by other people and other places. The reality, though, is that field experiments are not very common. They're difficult to perform. It's hard to have the power to randomly assign people in the real world to experimental conditions. Those that control the settings in which real groups operate, such as administrators and managers, often don't understand the method and don't want to cooperate—they just want to do the treatment, not test the treatments. Furthermore, controlling the situations such that the groups remain equivalent is also challenging. You know, something unintended by the researchers can happen to one of the IV groups but not the other (that's a "selection-history threat to internal validity"). They can share treatments, which is called a "diffusion of treatment threat" to internal validity. Dropout can be greater in one of the IV groups as opposed to another; that's called a "selection-mortality threat to internal validity."
[Music] A lot of studies in environmental psychology use correlational methods. These are non-experimental methods that can tell us if variables are related, or associated, that a change in one is systematically tied to a change in another. But they don't tell us what causes what. This method is not for that; that's the job of the experiment. So in correlational studies, variables are measured, not manipulated, and statistics are used to see how change in one is tied to the other. No IVs are manipulated in the correlational study.
Now many correlational studies use surveys to collect the data. So, for example, in one study I conducted interviews with people in a homeless shelter. Among other things, I was interested in the perceived amount of control they had in the shelter environment and how that was associated with learned helplessness. In another study, I used survey research to gather data on environmental concern and personal sustainability behaviors. Data from that study was used for a research publication that examined relationships between ethnicity and environmental concern. Correlational methods also include what are called "quasi-experiments." "Quasi," of course, means "sort of like," and they're called this because on the face of it they resemble an experiment, and if you don't know much about research methods, you might think it is an experiment. Now there are several types. The first one I'm going to mention is called a "non-equivalent control group design quasi-experiment." So, for instance, while we've been talking on the topic of noise, in the 1980s, a non-equivalent control group design was used to study noise. In this design, real-world groups that differ on a variable of interest are compared on another variable of interest. In this case, Sheldon Cohen and his colleagues hypothesized that long-term exposure to loud, uncontrollable noise would negatively affect children's persistence on learning tasks. The idea is that uncontrollable stressors can make people more passive by teaching them that they're helpless to control what happens to them; sort of like it shifts your locus of control to a more external one, which makes you more passive. They compared children at the four noisiest schools in the LAX airport flight corridor. They compared children at those schools with children at four schools that were matched on demographics but were not noisy. They tested the children in a noise-insulated trailer. They gave them a challenging puzzle, and then they measured persistence and success. In that study, in comparison to the children from the quiet schools, children from the noisy schools were more likely to give up before solving the puzzle. So again, this is an example of what's called a quasi-experiment. It is not a true experiment because the researchers did not randomly assign the children to the noisy schools versus the non-noisy schools, and while they can be reasonably sure that the differences in performance between the noisy and quiet schools were due to levels of noise because they made an effort to make their comparison groups, their "quasi-IV groups" if you will, as similar as possible such that they only differed on noise as best they could, anyway, it's still possible that there were other differences between the schools that caused the differences. And we call that a "selection threat to internal validity."
Quasi-experiments also include something called a "time-series design," and with that design, you have a series of measures of DVs that are collected over time. And you're looking for trends. You usually use government statistics. So, for example, imagine a graph that plots years' worth of monthly temperatures and crime rates. Now the biggest problem with time-series designs is a "history threat to internal validity." In other words, did something else happen during the time frame that might be responsible for the changes over time, like a new law or a new policy? In the case of crime, a police crackdown, or you know it could be the economy, right? These are other things that could explain differences over time in crime rate in addition to differences in temperature.
[Music] I want to say a little bit about surveys. Surveys are used when the only way to get the information is to ask. They can involve interviews. They can involve written surveys. They can be electronic surveys. Environmental psychologists use a lot of questionnaires and surveys. They measure things like environmental attitudes, beliefs, and behavior, and perceptions of environmental quality, among other things. Surveys and questionnaires are sometimes used to gather data for correlational studies, but they have other uses as well. When I was in graduate school, I did literally thousands of phone interviews with the citizens of West Covina, California, for their Department of Recreation to find out what local parks people frequented, what they did there, and what improvements they wanted to see there. I want to point out that survey and questionnaire design is much harder than it might look. A valid survey that reliably measures the relevant concepts takes a long time to construct. There are all kinds of tests along the way that help you ensure that you're reliably measuring what you intend to measure. I took entire courses in survey design when I was in graduate school. It's a complicated, thoughtful process. Any numbskull or knucklehead can write a daggone survey and slap it up on the old internet, but that doesn't mean it's a good survey—this is not for amateurs.
Now, in addition to the problem of poorly constructed surveys that don't accurately measure what is intended, there are also problems as far as the participants' memories go, and also with people giving socially desirable responses rather than truthful ones. There are ways to get around those problems, though, but again, it takes time, takes effort, takes expertise. Developing a measure that is psychometrically sound and is high in construct validity is no easy feat, and then on top of that you also have to think about sampling. Typically, you can't survey everyone in your population of interest, so usually you only survey a sample of them. But you want that sample to represent the population of interest—the ones that you want to apply your results to. Usually the best way to do this is with random or representative sampling. Random sampling means that every person in the population of interest has an equal chance of ending up in your sample, so you end up with a group of respondents (survey participants) that are a microcosm of the population of interest. There are all kinds of complicated things involved in this, like figuring out how large a sample you need to represent the population of interest. Representative sampling is carefully selecting respondents to mirror the qualities of the population of interest. So, for example, you might know that the population of interest has roughly this proportion of people from different age groups or different cultural groups, and you sample in such a way that you make sure that those proportions of different groups, different types of people in the population are represented in those same proportions in your sample. Sampling actually can get very complicated. It is hard to do well, but it is important because you can have a great measure, but if your sample doesn't represent who you want to apply it to, what do you really know? Sampling can be very complex, especially when a population is large and diverse. Then, once you draw your sample, you also have to work very hard to get everyone selected to participate, that is, to get a good "response rate." Otherwise, you can start off with a beautifully representative sample that ends up being non-representative because you couldn't get a lot of people to participate.
[Music] I want to talk next about naturalistic observation. I've been saying that environmental psychologists rely more on observational methods than most other research psychologists. Naturalistic observation is about observing what people do in real-world settings. Sometimes observation is used to collect dependent variable data, like in the recycling study when we drove around and looked to see who put their recyclables out on trash collection day, but that's not what we're talking about here so much. We're talking about observation that is not part of hypothesis testing. It's not measuring a dependent variable. There is no hypothesis. This is just about observing people in their environments and observing physical environments to see whether they meet people's needs. I'm going to tell you about two types, and the first type is called "behavioral mapping," and the second type is called "the examination of physical traces."
Behavioral mapping is a quantitative method. It involves a structured observational technique where trained observers code behaviors in an environment using a carefully designed checklist, and it's useful when you're considering making changes to a physical setting like a park, a playground, a library. You use observation to understand what people are doing in the environment, what they're using it for, and once you know that, then you can use that information to guide your renovations and other changes. Behavioral mapping has also been used in healthcare settings to see how much time is spent in treatment and rehabilitation activities. So let's say you're called in to do a behavioral mapping study at a psychiatric hospital. Administrators want to know how much time patients are spending in therapeutic activities. The first thing you would have to do is develop a complete list of what constitutes a therapeutic activity and have numerous different categories along with the behaviors that comprise each of those. Then you would train your observers so that they would all be in agreement about how to code a particular behavior, and then you'd also have to make sure that you systematically observed at different times of the day and that you represented all times of the week. Now, to use these methods properly, you have to meticulously develop your recording system, and another thing you have to do is carefully train your observers how to classify different instances, and usually if you're publishing this type of study, you have to prove that your raters, your observers, are all understanding the same behaviors in the same ways. You compute a special type of correlation to provide an index of inter-rater reliability; so that's considered very important to this method of behavioral mapping.
[Music] Now another interesting technique is what we call "the examination of physical traces." Physical traces are basically data that's left behind by people that were using a setting, sometimes by choice but often unintentional, like trash. But traces can also be conscious changes to the surroundings that people have made to make it better suit what they want to do there. Traces can also include markers of personal and social identities. So remember, physical traces are basically clues left behind by users of a setting—they can tell us things like who the people are that use the setting, what they do there, and whether the setting meets their needs, and also how they move through the space. Think about this as we talk about the four types of physical traces. Remember, physical traces give us clues about how people use the setting. "Erosions" can tell us what pieces or parts of the setting are used the most, and they can tell us how people move through the physical space. "Leftovers" are things left behind by users of the setting, and as far as interpreting them as clues, they show what people do there or what they did there, and sometimes they provide evidence of the ways that a setting doesn't meet people's needs. So, for example, overflowing trash cans or trash on the ground at a park can mean there aren't enough trash cans. When you go home today, take a look around for things left behind by the users of your home setting and see if you can tell what people did there while you were gone. "Adaptations of use" are a third type of physical trace. These represent users' modifications to the setting so that it can better serve their needs. These adaptations of use—these users' modifications to the settings so that it better serves their needs—can involve makeshift user solutions so that the setting works for them and things that they build or add or repurpose. They can also involve using existing setting elements in some creative and unique ways. Notice here in these pictures, for example, how people have created a bridge so that they can get over a creek. Pickleball users have taken PVC and created a place to put their rackets because the setting didn't allow for that when they wanted to play pickleball. So these adaptations of use can also show what users need from the setting and how the original design didn't meet their needs. Now I want to point out that erosions and adaptations of use sometimes coexist in the same physical trace. So here, for example, we can see a student-created shortcut from Cal Poly's PCV housing; that's an adaptation of use. It provides us with evidence that official paths didn't meet the students' needs, so they created their own shortcut, but it's also an erosion because see how it's worn away? That shows that this makeshift path is frequently used.
Displays of self and identity are a fourth type of physical trace. These are decorative items that people choose to display. They can tell us who the users are, what they like, what they value, and what groups they identify with. So, for example, if you came to my house, you would see a lot of decorative items that have nature themes. If you went to my son's house, you would see lots of items with cats on them. If you went to a person who was a devout Christian, you might see lots of crosses on the wall. If you went to a sports fan's house or to their room, you might see items representing their favorite sports team. People sometimes display items that represent who they are—you know, whether they're a proud Californian or what not. I should also point out that displays of self and identity also are sometimes territorial markers. People claim spaces, and they often use items to say "this territory belongs to me"—the setting, this room, this home, this desk, belongs to me. We'll talk more about that at the end of the course, but displays of self and identity are sometimes territorial markers; they are ways people claim physical spaces.
Thank you. Well, that concludes our brief lesson on research methods in environmental psychology. You'll learn more as we continue on our environmental psychology journey.
[Music]